Well, maybe I'm just stubborn, but I still don't think the bearings will be a problem. This unit came from an old powered wheelchair. I used to work on them for a while, and I can tell you that they get driven almost as much as they drive. They have to be built for both stop and go and long run operations. I've had my eye on these motors for quite some time, but they are usually encased in the transaxle and the output shaft of the actual motor is terribly small (like 1/4" X 1/2" total). This unit is almost the same internally, but has a nice 1" output shaft outside the gearbox. Oddly, it's suprisingly easy to turn despite the gearing.
As for the headlight trick: I tried that just for a joke and burned out the headlight - and my eyes - while turning the shaft by hand with a wrench(oops!).
I still see the main problem being getting it started. It's much like a cogging issue (I think?). It takes twice as much torque to start the motor as it does to keep it going. I thought of setting up a small wind sensor that would kick in a start current when the wind stayed above a certain speed for a certain amount of time, but that gets very complex very fast.
In the end, I think I'm just going to have to build a really large, really slow windmill design and see what happens. If I can keep the costs down, why not?